CTU Students’ Perceptions of Factors Influencing AI Chatbot Use in Learning English Skills

Abstract

"The rapid integration of artificial intelligence (AI) chatbots in higher education has created new opportunities for English language learning; however, the specific factors that shape students’ adoption of these tools remain insufficiently understood in the Vietnamese context. This study aimed to (1) identify which AI chatbots students use and the English skills they apply them to; (2) examine students’ perceptions of the factors influencing AI chatbot use; (3) determine whether these perceptions differ across gender, major, and cohort; and (4) assess which perceived factors significantly predict students’ adoption of AI chatbots in learning English skills. A mixed-methods design was employed, combining a questionnaire administered to 787 undergraduate students with semi-structured interviews conducted with four frequent chatbot users. Descriptive statistics revealed that chatbot use was highly concentrated on a small number of tools, with ChatGPT overwhelmingly dominant across all four English skills, particularly writing (87%). Students’ perceptions across all constructs (namely, Performance Expectancy, Effort Expectancy, Social Influence, Facilitating Conditions, Hedonic Motivation, Trust, and Habit) were generally positive. Among them, Performance Expectancy recorded the highest mean (M = 4.00), whereas Habit was the lowest (M = 3.32). ANOVA results indicated no significant gender differences, while significant differences were found between English-major and non-English-major students, as well as across cohorts, especially between Course 48 and Courses 50, 51. Exploratory Factor Analysis produced a refined five-factor structure: Habitual Trust, Ease of Use & Support, Hedonic Motivation, Social Influence, and Performance Expectancy. Multiple regression confirmed that all five factors significantly influenced chatbot use, explaining 70% of the variance (R² = .70), with Performance Expectancy and Habitual Trust being the strongest predictors (β = .43). Qualitative findings reinforced the quantitative results by showing that students experienced greater learning efficiency, faced prompting challenges, received social encouragement, engaged in self-directed learning, and developed selective yet habitual trust. Overall, the study provides a comprehensive understanding of students’ adoption of AI chatbots in English language learning."